Making Inference Across Mobilization and Influence Research: Comparing Top-Down and Bottom-Up Mapping of Interest Systems Joost Berkhout Department of Political Science University of Amsterdam Jan Beyers Department of Political Science University of Antwerp Caelesta Braun Department of Political Science Free University of Amsterdam Marcel Hanegraaff Department of Political Science University of Antwerp David Lowery Department of Political Science Pennsylvania State University This paper was prepared for Presentation at the ECPR General Conference in Glasgow, 2014 Abstract Students of mobilization and influence employ two quite different approaches to mapping interest systems. The first typically rely on a bottom-up mapping strategy in order to identify the total size and composition of the interest community. In contrast, students of influence have typically relied on a top-down mapping strategy in which the mapping of politically active organizations depends on samples of specific policy issues, usually those that are most prominent on policy agendas. But each domain of research also routinely uses data gathered to address research questions on mobilization or influence. How valid is this cross-domain use of large-n data on the politics of interest representation? In other words, are inferences based on a top-down mapping strategy generalizable to an interest system as whole and vice-versa? We address this question using unique datasets drawn from the INTEREURO project’s analysis of lobbying in the European Union (EU), datasets that employs both mapping strategies. Although, our analysis does not produce one simple answer to the question of comparability, our evidence shows that bottom-up and top-down sampling are considerably compatible and equivalent. We consider why this might be so and what research is needed to more fully assess the validity of cross-domain use of data generated from the two sampling strategies. Making Inference Across Mobilization and Influence Research: Comparing Top-Down and Bottom-Up Mapping of Interest Systems For most of the history of scholarship on the politics of interest representation, research was largely confined to only one or a few interest organizations. The reasons for this were obviously a lack of data on larger populations of organizations and their lobbying activities. Some notable scholarship was generated from this early empirical work, and it remains relevant to our understanding of interest politics to this day (Witko, in press). Still, following the strong advice of Baumgartner and Leech in their 1998 work, Basic Interests, the last two decades have seen a flowering of large-n research on interest organizations (see reviews in: Beyers et al. 2008; Bunea and Baumgartner 2014; Hojnacki et al 2013). Students of mobilization and influence have adopted research strategies that entail the mapping of large numbers of interest organizations active in politics and public policy so as to more fully account for the important contextual forces that govern both mobilization and influence.1 Students of mobilization and influence have, however, employed two quite different approaches to mapping interest systems. This is not surprising in that they are addressing somewhat different questions. The former are interested in who enters and leaves interest communities and the processes that govern their density and diversity. Answering such questions requires that we have a census of the total interest systems across time and/or space. Thus, mobilization scholars have typically relied on a bottomup mapping strategy in order to identify the total size and composition of the entire interest community. An early example of such research is Gray and Lowery’s (1996) analysis of U.S. state lobbying registration rolls in The Population Ecology of Interest 1 Please note that with ‘mobilization’ and ‘influence’ we refer to studies on research questions pertaining to, on the first, both mobilization (i.e the formation of a group and the decisions by existing organizations to become politically active) and organizational populations, and, on the latter, both political strategies and influence on public policy. We use these terms in a somewhat broader sense than elsewhere (e.g. Gray and Lowery 2004). 1 Representation. A bottom-up mapping strategy usually involves drawing samples from general registries of politically active organized interests, such as the U.S. state lobby registration rolls. In contrast, influence scholars are ultimately interested in how policy processes unfold, more precisely whether and how organized interests influence political and policy decisions. Addressing these questions requires that we attend first to the actual policy outcomes policymakers produce. Accordingly, students of influence have typically relied on a top-down mapping strategy in which samples of interest organizations are drawn from those active on specific policy issues, typically those that are most prominent on policy agendas. Perhaps the earliest example of such research is Heinz et al. (1993) analysis of the interests active in four prominent policy domains in the U.S. in The Hollow Core (recent examples are: Baumgartner et al. (2009), Beyers et al. (2014a, 2014b) and Mahoney (2008)). Importantly, neither mapping strategy in the now burgeoning literature relying on large-n research is more valid in and of itself; each is appropriate for its distinctive purpose of examining issues of mobilization or influence (Beyers et al. 2014b). Still, interest representation scholarship is not neatly divided between those who study influence and those who study mobilization. Students of each also routinely use data gathered to address issues of either mobilization or influence to address questions about issues better associated with the other. For example, in Interest Groups and Health Care Reform Across the United States, Gray, Lowery, and Benz (2013) employ data on the diversity of interests on U.S. state lobby registration rolls to assess their influence on the adoption of three health policies, implicitly assuming that the diversity of total lobby registrations is isomorphic with the diversity of interests actually lobbying on the policies. Similarly, while Klüver’s Lobbying in the European Union, is largely devoted to the analysis of influence based on a sample of organization submissions on 2 proposals under consideration by the European Commission (EC), it also addresses broader issue of mobilization and the diversity of the EU interest system (2013, 138144) using the same data. Extracting broader generalizations on these latter questions from data on the explicit activities of organized interests requires, of course, an assumption that the diversity of interests active on a sample of on-going policy decisions map in an isomorphic manner on the diversity of the larger interest system. Clearly, then, it is not uncommon for data gathered via one mapping strategy and designed to test hypotheses about either mobilization or influence to be used to say something about issues more appropriately in the other domain. How valid is this cross-domain use of large-n data on the politics of interest representation? Can data gathered on the diversity of interest organizations from a bottom-up mapping of the full interest population be validly employed to address issues of influence on important political and policy decisions? And should we be concerned about the use of data gathered on the lobbying activities of organized interests on a sample of important issues to tell us something about overall patterns of mobilization within political systems? We address these questions using data drawn from the INTEREURO project’s analysis of lobbying in the EU (Beyers et al. 2014), a project that employed both mapping strategies. We first outline the many reasons to suspect that the two mapping strategies should indeed generate quite different distributions of interest organizations. Then, we compare the distributions of interests generated by the two INTEREURO sampling strategies across six variables generated for both: the level of mobilization of the organization, their organization structures, the geographic location of their headquarters, the type of actor they are, and either the economic or social sector from which they are drawn. In a second empirical portion of the paper, we assess the degree of overlap in the two mapping strategies in terms of identifying individual 3 interest organizations. Importantly, although our analysis does not produce one simple answer to the question of comparability, our evidence shows that bottom-up and topdown sampling are considerably compatible and equivalent. In the conclusion, we consider why this might be so and what research is needed to more fully assess the validity of cross-domain use of data generated from the two sampling strategies. The Case for Incongruence in Mapping Strategies As noted earlier, scholarship on the politics of interest representation is not so neatly divided between those interested in mobilization and those concerned about influence. Many scholars routinely move between the domains, and often bring their data with them. But to better understand why such cross-use of data generated from the two mapping strategies might be of concern, we might adopt a hard critical stance from one or the other domain and assess why it is problematic to employ data generated from the preferred mapping strategy employed by the other domain. There are several reasons why students primarily interested in mobilization might be concerned about comments made on those topics by scholars using mostly data generated from top-down samples of organizations considered either most active in general or most active on a sample of on-going and often important issues before a government of the type employed by Heinz (1993), Klüver (2013), Yackee (2006), Beyers et al. (2014b), Mahoney (2008) and Baumgartner et al. (2009). First, and perhaps most importantly, a focus on policy decisions is simply besides the point from the perspective of many scholars given that organizations might mobilize (i.e. develop political activities or monitoring the policy process) for any number of reasons beyond actually seeking influence on specific policy choices. Indeed, joining a lobbying community may be more about mobilizing members or securing finance than about 4 policy influence per se (Baumgartner et al 2009; Braun 2014; Lowery 2007; Hanegraaff 2014). Attending to these concerns may lead to little or no explicit effort to actually influence policy decisions, and organizations motivated by such concerns would be missed or at least underrepresented in samples drawn from those actively seeking influence on specific policy decisions. Second, even when organized interests are primarily concerned about influencing public policy, the policy process is a long and winding one, running from efforts to get any kind of hearing to final and authoritative policy choices on concrete proposals. And importantly, most interest organizations fail to even get their preferred issues on the table (Baumgartner et al. 2009). Such organizations would be absent from samples generated using a top-down strategy since these are typically attentive to lobbying activity on issues that have already secured space on the policy agenda in their form of a concrete policy proposal. Thus, active proposals are likely a biased sample of all issues considered by policy makers, and this, in turn, is likely to generate a biased sample of interest organizations compared to all who might be seeking policy change. Indeed, topdown mapping often assumes, if implicitly, that such activity is in response to demand function generated by the introduction of a proposal (Leech et al. 2005). The inadequacy of this assumption is fully evident in lobbying in the EU where there is no clear pattern in the timing of the appearance of a specific policy proposal and lobbying activity by interest organizations (Toshkov et al. 2013). Much of lobbying – if often unsuccessful – occurs long before a specific proposal appears on policy agendas, which suggests that samples of interest organizations drawn from lobbying activity on specific and often important policy proposals may miss a great deal of lobbying and many 5 interest organizations.2 Third, to mobilization scholars, lobbying involves much more than trying to secure a yes or no vote on a policy proposal. Indeed, much of “lobbying” is merely occasional policy monitoring or dipping into the policy process on only an episodic basis. Indeed, interest communities are highly dynamic, with many organizations appearing only occasionally and then disappearing from the interest population (Berkhout & Lowery 2011; Anderson et al. 2004). But this research has also shown that interest communities tend to have a number of persistent members – the “old bulls” that are routinely present within interest systems. It would seem plausible that such “old bulls” or “permanent residents” are overrepresented in samples generated from topdown mapping strategies and that the “mayflies” or “tourists” are seriously underrepresented, as are those who engaged in only more passive forms of policy monitoring. Yet, students of mobilization, given their concerns about patterns of exit from and entry into interest communities are as equally attentive to both the old bulls and the mayflies; each is equally relevant to assessing the degree to which interest systems are open and unbiased. But a sample based on influence activity at the final stage of the policy process may miss many of the latter. And fourth, there is a long history of suspicion of bias in interest systems in research on interest representation. Interest systems are often viewed as overrepresenting a few large interests, typically business interests (Schattschneider 1960; Schlozman et al. 2012). The diversity of interest communities is, of course, a major topic of concern to mobilization scholars. But there is also an implicit assumption in the 2 Some top-down samples (like that generated by Baumgartner et al. 2009) are somewhat less subject to this problem given that they start with the proposals organizations are working on rather than sampling interest organizations from those working on important issues at the final stage of the policy process. Also the INTEREURO-project addresses this issue by adopting a stratified sampling strategy which ensures that both salient and less salient issues are included in the study (see below). 6 literature that interest communities may become even more biased as policy proposals move toward final decisions. That is, contrary voices, through a process that might be labeled an iron law of influence concentration, may become increasingly less active, have less access, and have less influence as proposals move through the policy process, and the scope of the conflict becomes narrower. As noted by Schattschneider (1960, 34) ‘scope and bias are aspects of the same tendency’, and the narrower the scope of conflict, the stronger the bias of the interests involved. From the perspective of mobilization scholars, the potential of increasingly less access is especially problematic. If interest systems become more concentrated and biased as a proposal moves through the policy process, then samples drawn via top-down mapping may exclude many organized interests with a legitimate interest in a policy proposal.3 In sum, then, there are many reasons for mobilization scholars to suspect that data generated by top-down mapping provides a valid sample of interest organizations that can be used to address mobilization-related questions. Conversely, there are several reasons why students primarily interested in influence might be suspicious of comments made on that topic by scholars using data generated from bottom-up samples of organizations based on lobby registration rolls or some other census of the entire lobbying community of the type employed by Gray and Lowery (1996); Gray et al. (2013), Mahoney (2008), Hanegraaff (2014), Halpin and Binderkrantz (2011), Berkhout (2010), and Poppelaars (2009a;b). Many of these are the obverses of those raised against using top-down samples by mobilization scholars. That is, if the focus is really on the influence of organized interests on final policy 3 This Schattschneiderian expectation (the further along the policy process, the narrower the scope of conflict, the lower the diversity) runs counter, of course, to the other Schattschneiderian hypothesis that the scope of conflict is expanded in the process from agenda-setting onwards and through the involvement of public authority and, most notably, ‘public conflicts’ in the legislature (Schattschneider 1960, 40). 7 decisions, then those organizations lobby for some other reason than securing influence – the “visitors” or “mayflies”, those denied any real access to decision makers, and those who fail to get their proposals on the policy agenda – are less relevant for analyzing how concrete policy processes unfold. Thus, data from samples generated from a bottom-up sampling strategy may be contaminated with the counting of many interest organizations that either not seeking influence or are prevented from doing so. How plausible is such contamination? On first face, it would seem to be significant. It seems likely, for example, that the activities of membership groups and non-profits are more heavily motivated by concerns of mobilization and finance than are institutions. This implies that such organizations would be over-represented in bottomup samples, but less visible in top-down samples. It also seems that large businesses and business associations would be more likely to maintain a constant presence in lobbying communities – and be part of the set of “permanent residents” – than smaller business interests. It seems plausible – at least from a Schattschneiderian perspective in which ‘pressure politics is a selective process ill designed to serve diffuse interests’ (1960, 35) – that access to policymakers during the final decision-stage is easier to obtain for wealthy business interests than for NGOs and other public interest groups; they are more likely among the “old bulls”. And if the policy proposals of some types of organizations receive a more ready hearing and, thus, space on the policy agenda than those of others (Schlozman et al. 2012), then the distributions of organizations at the table when making authoritative decisions on public policy may be skewed. It seems, then, that is plenty of reason to suspect the data generated from a bottom-up census of the interest population will represent a biased and unrealistic representation of those truly active at the final stage of the policy process. But even if the distribution of interest organizations generated via a top-down 8 sample of active interest organization faithfully mirrored those produced via a bottomup census of interest organizations, there remains a more serious problem for influence research relying a bottom-up census of interest organizations. This problem, if not perhaps an inherent one, lies in the nature of the data typically generated by the two research strategies. That is, because bottom-up research strategies typically attempt to map whole interest communities, they are severely limited in assessing the amounts or kinds of activities organizations undertake beyond merely showing up somewhere in the policy process. To use such census data to study influence, one has to basically assume that all interest organizations do more or less the same thing once present. In contrast, top-down samples, because they work with a smaller sample of issues, typically do much more than merely registering presence or absence in the policy arena. They assess lobbying strategies and tactics, evaluate comments and statements used to frame choices, and, among many other things, look at the frequency and locus of explicit lobby activity. These kinds of data allow scholars to weigh interest organizations in terms of the intensity of their efforts to influence decisions, something that cannot be done with a simple count of interest organizations. For these reasons, it would seem doubtful that a census of organized interests used for mobilization research could validly serve as a quick and ready substitute for a top-down sample. Is, then, the prospect of cross-domain use of datasets generated using top-down and bottom-up generated lists of interest organizations really so bleak in terms of inferential validity as the preceding arguments suggest? Perhaps not. Despite their many differences, most interest organizations employ a fairly standard, common repertoire of activities once they join interest communities (Walker 1991; Baumgartner et al. 2009, 149-150). This implies that weights are not needed for levels or intensity of lobbying by organized interests. And given the high frequency of entry into both U.S. 9 and European interest systems (Berkhout and Lowery 2011; Anderson et al. 2004), both systems remain fairly open to participation on the part of interest organizations of all types. Indeed, there is in the end nearly always two sides to policy debates, thereby insuring that one type of interests does not fully dominate policy discussions (Baumgartner et al. 2009). Further, prior research suggests the vast bulk of lobbying is done on only a few proposals considered by governments (Baumgartner & Leech 2001; Halpin 2011; De Bruycker & Beyers 2014). If this is true, then a top-down sample of interests active on a sample of major proposals should be similar to one drawn from the full interest population. And perhaps most importantly of all, research on influence topics using bottom-up or population-level data on interest organizations has not generated findings notably different from those relying on a true top-down sample of interest organization activity, and research on mobilization questions based on a topdown sample of active interest organizations has not produced findings that are discernibly different from those using a bottom-up census. For instance, relying on nonissue focused samples Binderkrantz and colleagues (2014) demonstrate that interest group strategies are broadly consistent with studies using policy-centred samples. Still, we cannot know if this is a happy accident or if it masks a more fundamental inference problem given that there is to date no research that has simultaneously employed both types of data so that we can meaningfully compare the descriptions of a common interest community. Comparing the Two Research Strategies We provide such a comparison in the remainder of this paper. Our data were generated from the INTEREURO Project (Beyers et al. 2014a; 2014b), which employed 10 both types of mapping strategies in its analysis of lobbying in the EU.4 We first describe the two mapping strategies and the data. We then turn to looking at both the overall distributions of organizations generated by each on six variables and the overlap in the individual organizations identified via each strategy. The INTEREURO Bottom Up Mapping The ideal bottom-up data source includes all organizations in a given polity that engage with public policy. In conceptual terms, each organization showing some activity in relation to the policy process is defined as interest organization, including individual institutions such as banks, hospitals or regional authorities. An interest organization enters the population when it shows such an activity and exits the population when it is no longer politically active in the polity under study. Remember that ‘political interest’ is commonly part of any definition of ‘interest groups’ (e.g. Beyers et al. 2008; Salisbury, 1994). Here we employ a behavioral definition of interest organizations rather than an organizational-structural one. In the latter, and in contrast to our approach, a bottom-up population would exclude individual institutions but include all collective, non-party organizations with ‘political’ goals, regardless of whether these goals are observably articulated in the policy process (see discussion in: Halpin & Jordan 2012, Jordan et al. 2004). The organizations in the doorpass register of the European Parliament (EP) between 2008 and 2010 form our bottom-up population.5 The EP issues entry passes to persons who wish to enter the Parliamentary premises frequently and engage in 4 The INTEREURO data are or will shortly be available at www.intereuro.eu. The list includes data on 23012 entry passes of persons affiliated to 3704 different organizations and was provided by the Parliamentary Information Systems Unit (SIP) of the EP in May 2011. The earliest passes in that version are from early 2004 and the newest are from April 2011. The list is governed by Rules of Procedure of the EP (Title 1, Chapter 1, Rule 11, 5-11, also see Annex IX). 5 11 activities ‘carried out with the objective of directly or indirectly influencing the formulation or implementation of policy and the decision-making processes of the EU institutions’ (EP rules, Annex IX(8)). Entry passes are valid for one year and may be renewed. This ensures that data is relatively up-to-date. We take a two-year time period in order to match the bottom-up generated population with the top-down generated population based on policy proposals drawn from the period 2008-2010, see below. The precise rules governing the issuing of passes have remained relatively constant over the time period studied (but note that, for instance, the introduction ‘express’ passes mid2006 led to a decline in the aggregate number of passes). We have several reasons for relying on this register rather than other data sources. First, in the context of the EU institutions, the EP is designed to be the most open to organized interests. In the end, it is the role of Members of the European Parliament (MEPs) to represent interests and political positions. They consequently have a reason to converse with various organized interests, and, in light of the importance of legislative decisions, various interest groups also would like their views heard by MEPs. The EP compared with the EC and Council should have the lowest threshold of policy participation for interest groups and it shows at least some openness regarding these practices, as noted by Balme and Chabanet (2008 216) ‘the creation of a system of regulation and supervision of lobbying bespeaks of a political will (…) specific to the European Parliament’. The consultations of the EC have a similarly low threshold of participation but only occur on specific, and certainly not all, proposals. Organizations with some interest in EU policies are likely to, at some point over a longer time period, want to, and are welcome to, enter the building of the EP. This makes the register is a good bottom-up population data source compared to data sources (for instance those associated with other EU institutions or published by private companies).. 12 Second, registers of this type are common in interest representation studies. The register of the EP itself has previously been used for various research purposes (Toshkov et al. 2012, Wonka et al. 2010). The use of lobby registers is also relatively common outside the EU case. Most notably, Gray and Lowery state that in the US case ‘the most valid indicator of broadscale political activity now available is provided by lobby registration rolls’ (Gray & Lowery 1996, 7). The Lobbyistenregister of the German Bundestag has been used for several surveys (Dür & Mateo 2013, 679, Jentges et al. 2012, Klüver 2012). Even though the precise requirements vary per institution, the common usage of such registers still provides some confidence in the relative quality of the data. Third, alternative data sources are however not better. To start, the Transparancy Register (or ‘Register of Interest Representatives’), in place as successor of the official list of the EC (CONECCS) since mid-2009 and as joint EC-EP list since 2011, only very recently reached a level of coverage that makes it potentially useful for academic purposes. The time-period consequently does not match with our top-down population data. The EP register, in place since 1996 (of which we cover 2008-2010), runs over a longer time period, and most importantly during the time period in which we select organizations based on the top-down sampling strategy. Further note that currently all organizations in the EP register are expected, when possible, to register in the Transparency Register.6 By using the EP list we effective rely on a sub-sample of the 6 But note that there may be a number of organizations in the EP list that are not in the Transparency register because of the ‘scope conditions’ of the latter. EP administrators have quite some leeway in issuing badges: ‘Rule 11 (5) states: Such badges may be issued to: - persons whose names appear in the transparency register or who represent or work for organizations whose names appear therein, although registration shall not confer an automatic right to such a badge;- persons who wish to enter Parliament’s premises frequently, but who do not fall within the scope of the agreement on the establishment of a transparency register(3); - Members’ local assistants and persons assisting Members of the European Economic and Social Committee and the Committee of the Regions.’ Transparency campaign groups, most notably ALTER-EU, are especially concerned about the exclusion of various types of legal advice, for instance, Annex IX, 10(a) excludes from the register any ‘company and its advisers [who]are involved as a 13 Transparency Register. This is in line with the advice of Greenwood and Dreger (2013, 159) that ‘for research use’ a selection filter is needed ‘to make the data less prone to quirky outliers’.7 Greenwood and Dreger (2013, 150) note that in January 2013 1179 out of 5494 organizations from the Transparancy Register hold access passes to the EP. The overlap between our EP register data and the Transparency register is likely to be bigger because we rely on a two-year period rather than a snapshot. EP passholdership is also likely to distribute relatively evenly over categories of registration in the Transparency register. That is, Greenwood and Dreger (2013, 149, 152) report only a somewhat larger proportion of EP passholders among business associations (19%) than among non-governmental organizations (15%).8 Another alternative data source would be any one of the several public affairs directories. These, most notably the European Public Affairs Directory published by Landmarks (since 2011 Dod’s), have been used in a couple studies (e.g. Sorurbakhsh 2014; Berkhout & Lowery 2010; Wonka et al. 2011). Such commercial directories are marketed to the ‘Brussels community’ of policy makers, stakeholders and civil servants. Their registration practices seem to favor interests with permanent and relatively central presence in Brussels and that are ‘collectively’ organized. At the same time, such directories seek the broadest possible coverage to market the large number of entries in party in a specific legal or administrative case or proceeding, any activity relating directly thereto which does not seek as such to change the existing legal framework’. 7 Please note that they recommend to select only organizations with a ‘European interest’ or a Brussels address. As said, the main benefit of the EP pass criterion is that it indicates policy engagement. 8 Greenwood and Dreger (2013, 149-152) find the following distribution of EP passholders across the registration categories: Transparency Register category n passholders (I) Professional consultancies/law firms/self-employed consultants (II) In-house lobbyists and trade/professional associations (III) Non-governmental organizations (IV) Think tanks, research and academic institutions (V) Organizations representing churches and religious communities (VI) Organizations representing local, regional and municipal authorities, and other public authorities 14 124 541 193 36 10 47 N in Transparancy Register 620 2816 1304 411 37 306 % 20% 19% 15% 9% 27% 15% their directory (see Berkhout & Lowery 2008). We do not use any of these directories because the time period reliably covered does not match the time period of our top-down sample. That is, around 2010 the European Public Affairs Directory was just ‘in transition’ between the publishers Landmarks and Dod’s. Other directories, such as stakeholder.eu, the Europa-OECKL or Euroconfidentiel, were not yet in print (or just out of print). Previous studies show some overlap between Landmarks and EP register: Wonka et al. (2010, 466) removed 487 organizations as duplicates arising from the addition of the EP register (n=1534, version April 2008) to the combined CONECCS-Landmarks list (CONECCS: n=749, Landmarks: n=2522, overlap: n=489). The inclusion of organizations of the EP register over a longer time-period, as we do, is very likely to produce a larger overlap with the Landmarks PA directory. However, there probably remains a substantial set of organizations that appears in the Public Affairs Directory but not in the EP register, even over the longer term because the PA directory also lists organizations that take a more distant approach to policy activities and are mainly working on ‘interest aggregation’ activities such as annual conferences, networking among member-associations and so on. Such organizations may never lobby the EP. Future studies probably benefit from combining the EP register data with directories such as the Public Affairs directory or the Transparency Register. The INTEREURO Top Down Mapping The starting point for the INTEREURO top-down mapping was a list of all proposals for generally binding EU law, i.e. directives (N=144) and regulations (N=459), adopted by the EC between 1 January 2008 and 31 December 2010 (Beyers et al. 2014a). One reason for selecting a particular time slot was to maximize the chance of 15 finding interview partners in the Commission as well as among organized interests. We also wanted to be reasonably certain that at the time when we did our interviews a decision would have been taken on the proposal (so no proposals that were too recent could be included; see also Beyers et al. 2014c). After eliminating a few proposals concerning the remuneration of and/or pensions for Commission officials, the internal management of the EU institutions and codification proposals, the list contained 111 proposals for directives and 427 proposals for regulations.9 Importantly is that we stratified the sample of legislative proposals according to public saliency, which increased the likelihood to sample highly salient cases. We did this by checking the media coverage of all proposals in five sources, namely Agence Europe, European Voice, the Financial Times, the Frankfurter Allgemeine Zeitung, and Le Monde. About two thirds of the proposals for directives and sixty per cent of the proposals for regulations were mentioned in at least one media source. We found most media hits in Agence Europe, and fewest hits in Le Monde. The resulting sample includes 48 proposals for directives and 38 proposals for regulations that were mentioned in at least two of these media sources. We opted for this rather low threshold of two media hits in order to allow for a substantial variation in salience across proposals in the sample. Moreover, to add additional variation in terms of salience, we added a randomly selected set of ten proposals for directives and nine proposals for regulations that did not meet the media coverage criterion. Finally, we added all proposals for directives and regulations that had not made it into the sample following our strategy, but for which public consultations had been held and consultation documents are available (Klüver et 9 We decided to sample directives and regulations separately, as a simple random sample would have made us select very few directives (about four regulations for each directive). Having a large enough number of directives in the sample was important as we wanted to see whether there are any differences in terms of lobbying between regulations and directives, and because for directives we can also study lobbying during the national transposition process. 16 al. 2014). We did this for pragmatic reasons, as we wanted to benefit from the additional data that is available for consultation cases. This process resulted in a sample of 125 proposals, which breaks down to 64 proposals for directives and 61 proposals for regulations. Our next steps were a) a thorough media analysis of all these proposals and b) interviewing projects with experts in the EC, the EP and interest group officials. The interviewing projects took the set of 125 cases as a starting point. Table 1 around here Via an online search in the electronic archives of five media outlets (European Voice, EurActiv, Agence Europe, Le Monde and Financial Times) we mapped the complete media coverage related to the 125 legislative cases. Important is that only coverage that could be directly connected to the EC proposal was kept; coverage that was only vaguely related to the subject matter of the proposal was not archived. We adopted this rather strict rule as we wanted to concentrate on coverage related to legislative issues, and not necessarily the coverage on some more general theme indirectly or vaguely related to the legislative case. In a next step we identified all stakeholders (public and private) and stored all the statements these actors made in a separate database. Statements are quotes made by interest group officials (for instance in an interview with a journalist) that can be directly linked to one of the sampled proposals (all these statements were coded as well; see De Bruycker & Beyers 2014). This part of the mapping implies that we can make a distinction between actors that are only vaguely mentioned (n=288) from those that actively contribute to some media debate (n=379). 17 For each of the 125 sampled pieces of legislation the responsible policy unit and policy expert from the Commission was identified and invited for an expert interview.10 In 29 cases potential respondents refused to be interviewed or officials argued that they did not remember enough about the specific case. For 26 cases very short preliminary interviews took place, sometimes over the phone, which showed that no or very little lobbying took place. In the end, this led to 70 legislative cases for which lobbying was confirmed and extensive face-to-face interviews with EC-experts were conducted. During these interviews, the expert was asked to list all relevant stakeholders that were actively involved in the specific legislative case, and to identify key issues, the main conflict dimensions and policy positions that stakeholders had adopted. In this way we identified 658 interest groups. In the next stage of the interview project, following the interviews with 95 EC-experts, we conducted 141 interviews with interest group officials. In the selection of respondents we relied on evidence collected through the media analysis and the EC-expert interviews. One important part of the interview consisted of a detailed screening and characterization of all the actors (for instance identifying who was very active, who were coalitions leaders, important providers of expertise) we identified in the previous steps, plus adding actors that remained unidentified. The interviews with group officials started with the 70 proposals on which EC-expert interviews took place, but we added 15 proposals where we encountered considerable media attention in terms of interest groups that made public statements on this case. In this way we identified an additional set of 291 organizations. Data collection Interest groups from the EP register as well as the groups that were identified 10 This work was largely conducted by the Austrian INTEREURO team. 18 through the top-down mapping were coded on a number of different variables (see codebook www.intereuro.be). Information was coded mainly on the basis of organizations’ websites, and, sometimes, also on the basis of annual reports, information submitted to the transparency register, and other sources of publicly available information. Several student-coders collected the data under supervision of a postdoctoral researcher. They were trained in three distinct groups of four students, held regular meetings to cross-verify the treatment of specific cases and inter-coder reliability checks were satisfactory. The key benefit of our reliance on online, publicly available sources rather than surveys is that it provides us with a very high ‘response rate’. Only a very small proportion of defunct or marginal organizations could not be found online. Our approach does not suffer from major selection bias, as may to a certain extent be the case for survey research, and to a lesser extent face-to-face interviews. Yet there are a couple of limitations associated with the use of website information. First, , we face problems of missing data, in particular for variables measuring organizational resources and staff size the response rate is much lower than fifty percent. And even for those organizations where we found information on staff size, it is surely impossible to differentiate, on the basis of organizational websites, between ‘lobby’ staff and other staff. This validity problem leads us to disregard this part of the data collected. Second, actual, offline policy-oriented activities of organizations is sometimes poorly presented. This makes it very difficult to validly identify the policy interests of organizations. For that reason we refrained from classifying organizations into the main categories of the Policy Agendas Project (see: comparativeagendas.org), although this is a relatively common type of data collection strategy is such research projects (e.g. Jordan et al. 2012; Johnson 2013). Much of this information could be 19 obtained through interviews with the ‘top-down’ sample, but as this type of data is lacking for the whole population we will not use it for this paper. For each of the variables presented below, these validity problems do not occur or are not so severe. For this paper we rely on our coding of the governance level at which an organization is mobilized, their organization structures, the geographic location of their headquarters, the organizational form, and either the economic or social sector from which they are drawn. Empirical Analysis It is clear from Figure 1 that the top-down sampling of organized interests active on EU-legislative processes identifies many organizations not included in what is the most comprehensive available bottom-up census of those lobbying in Brussels. While the top down sample consists of 935 organizations, 588 (or 63 percent) are not part of the bottom-up population. Moreover, only 347 appear in both interest populations, which constitutes 37 percent of the top-down population and only 15 percent of the bottom-up population. There are many potential reasons for this. Most prominently, both types of mapping strategies are incomplete. Top-down sampling may pick-up organizations, for instance through media reports, that do not actually lobby. Even experts may under- or overestimate the lobbying activities of some organizations and any bottom-up population census is likely to be incomplete.11 We will discuss these reasons and others much more thoroughly in the conclusion. For now, however, it is worth adding these sampling problems to the list of more substantive reasons why 11 Even the quite rigorous lobby registration of the U.S. national government under the Lobby Disclosure Act of 1993 is incomplete via exclusion of organizations spending less than a fixed dollar threshold, those that lobby only via the media, and those active only in electoral campaigns via PACs. This problem is even more sever in the EU where we lack a rigorous registration system and where, due to the multi-layered system of the EU, boundaries of the interest groups system are even more difficult to draw. 20 bottom-up and top-down mapping strategies may generate different distributions of interest organizations. Figure 1 about here We compare the distributions of interests generated by the two INTEREURO sampling strategies, as well as the individual cases they identify, across six variables generated for both: the level of mobilization of the organization, their organization structures, the geographic location of their headquarters, the type of actor they are, and either the economic or social sector from which they are drawn. We will not say much about the individual values of these variables, although these will be apparent in our discussion of the data, since the issue here is not measurement validity per se, but the comparability of their values across two different sampling strategies. Nonetheless, these variables are of long-standing concern in both the literature on organized interests and among those concerned about interest representation in the EU. We should note, however, that not all interest organizations were coded on the last two variables: the ISIC codes of economic organizations and the social codes for organizations lacking an ISIC code; all organizations were coded by one or the other of the variables. We will also compare the balance of these two different types of organizations across the two mapping strategies in the INTEREURO dataset. To assess the comparability of the data generated by the two samples, we first examine the overall distributions of the six variables. The full tables on these distributions are presented in appendixes 1 through 6 where the five pairs of columns present frequencies of the six tables for five subsets of the total dataset – 1.) the cases uniquely generated by the bottom-up sample (not recorded in the top-down sample), 2.) 21 the cases included in the total bottom-up sample, 3.) all cases identified by both mapping strategies (after eliminating duplicates), 4.) the cases included in the total topdown sample, and 5.) the cases uniquely identified by the top-down sample (not recorded in the bottom-up sample).12 Frequency distributions drawn from these tables are presented in Figures 2 through 7 for the same five sets of cases. Figure 2-7 about here There are some differences evident across the distributions generated by the mapping strategies. These are perhaps best evident via comparing the second and fourth bars of the figures – to total bottom-up and total top-down datasets, respectively, recognizing that these include some of the same interest organizations. In Figure 2 on level of mobilization, for example, non-EU multinationals are somewhat better represented in the bottom-up sample (37.40 percent) than in the top-down sample (31.53) percent, suggesting that they are more passive or are more likely to restrict their activities to policy monitoring only (and to not lobby that many legislative cases). In figure 4 on the geographic headquarters of interest organizations, U.S. and Canadian interests are less well represented in the total bottom-up sample (6.24 percent) compared to the top-down sample (12.55), the reverse of what is found for Belgianheadquartered interests – 29.35 and 22.64 percent, respectively. In Figure 6 on the ISIC codes for economic sector, finance, insurance, and real estate interests are less well represented in the total bottom-up sample (7.77 percent) compared to the top-down sample (21.50). In Figure 7 on the social codes for interest organizations lacking an ISIC code, employer and producer groups interests are less well represented in the total 12 The numbers of observations in the appendices tables differ among each other because of unique sets of missing values on the six variables. 22 bottom-up sample (38.69 percent) compared to the top-down sample (49.52), the same that is true of environmental interests – 9.95 and 18.33 percent, respectively. Thus, the two mapping strategies do not generate precisely the same distributions of interests across the six variables. Figure 8 about here At the same time, however, the primary message of Figures 2 through 7 is that these distributions of interests generated by the two mapping strategies are remarkably similar in their overall distributions. This is evident in Figure 8, which presents correlation coefficients for the distributions in the second and fourth columns of Figures 2 through 7. All of the coefficients are 0.942 or higher except for the distributions for the primary social code of economic organizations, which was 0.800 (see also De Bruycker & Beyers 2014). Another way to see this overall similarity is in terms of what is perhaps the primary concern of scholars concerned about interest system diversity and bias – the balance of economic and social organizations. As noted earlier, interest organizations were coded either by an ISIC code (Figure 6) or a social code for organizations lacking an ISIC code (Figure 7). In the total bottom-up sample, 68.85 percent of the cases were assigned an ISIC-code, nearly identical to the 69.83 percent that were so assigned in the total top-down sample. Conversely, 31.15 percent of the cases in the total bottom-up sample were assigned a social code, which is quite similar to the 30.17 percent so assigned in the total top-down sample.13 Thus, despite our expectations of sharp differences in the tallies of interest organizations across the two mapping strategies, it seems as if the list of organizations most active on major policy 13 While we label those lacking an ISIC code social groups, such organizations, as is evident in our discussion of figure 7, do include economic organizations, most notably employer and producer groups. 23 issues in the EU (the top-down sample) appears not especially dissimilar from what might be generated from a random sample of all of the interest organizations lobbying the EU as generated via the best available bottom-up sample of EU interest organizations. While the distributions of the bottom-up and top-down mapping efforts across the six variables of interest are quite similar, do they tally the same individual organizations? This is certainly not the case as was evident in figure 1; the top-down sample of organizations was not fully nested in the bottom-up population census. This is typical across the six variables we have examined as well, as seen in appendix 7.14 The most telling issue here concerns how many of the organizations tapped by the top-down sample of those actively lobbying of the prominent issues before the EU were included in the bottom-up census of the population of interest lobbying the EU. These proportions are reported in figure 9 for the six variables of interest. The majority of the lists of interest organizations tallied by the top-down mapping of those active in lobbying the prominent issues before the EU were also included in the bottom-up census of all organizations active in lobbying the EU for only one of the six variables of interest: he primary social coding variable (52.381 percent). On all of the other variables, a majority of cases were not included in the bottom-up sample. Conclusion This leaves us with something of a puzzle. It seems the distributions of the observations generated by bottom-up and top-down mapping procedures generate quite similar distributions on our six variables. This is good news for scholars studying 14 Again, the number of observations in the this table differ from each other and that reported for figure 1 due to missing data on the individual variables. 24 the politics of interest representation in that we should feel more free to generalize about broader mobilization issues using data generated from sample of interest organizations active on major policy issues. One important condition is, however, that scholars are conscious about the various potential sources of bias and variation when they design studies with bottom-up and top-down samples. And we should feel more willing to employ bottom-up population-level data in analyzing research questions related to the relative influence of different kinds of interests on specific policy issues. At the same time, however, we have seen that the two mapping strategies do not capture the same individual organizations. More specifically, the top-down sample of organizations is not subsumed by the bottom-up census of interest organizations generated through the bottom-up mapping procedure. This means that we have to remain careful and cannot unambiguously attribute any observed influence effects to specific constellations of interest organizations when using bottom-up data to identify the relative strengths of different types of competing interest organizations. In particular it is important to consider why small fluctuations emerge due to some policy agenda effects (e.g. the large number of banking and financial interests). The most likely explanation for why the organizations identified by the top-down mapping process are not fully captured by the bottom-up census lies in the inherent inadequacies of both types of samples. This implies, of course, that their combination, while excluding duplicate entries, is a useful approach to addressing their inadequacies, although only the INTEREURO project has been able to combine and compare both mapping strategies to date. The inadequacies of the bottom-up mapping process are more obvious than those of the top-down strategy. Simply put, the EU does not have anything like a comprehensive and rigorous registration system. Moreover, the multilayered structure and transnational nature of the EU makes that the boundaries of its 25 interest group system are difficult to establish. The door pass data we have used in the INTEREURO project is, we believe, the best, most comprehensive data now available. But it is incomplete. In particular, it is less likely than a full registration system that tallies lobbying in all EU institutions – as is done under the Lobbying Disclosure Act of 1993 in the U.S – would be to capture influence at the earlier agenda-setting stage of the policy process, which is likely to be centered on the EC. And quite simply, those active at that stage may not be active at later stages of the policy process that include decisions taken in the EP and the Council. In particular, advocates that support the Commission and that face strong opposition in the EP and/or the Council might be reluctant to put much pressure on policymakers in the EP as this may simply attract the attention of their opponents. At the same time, however, the top-down procedure might capture interest organizations that do not very actively lobby on specific policy proposals or that for whom lobbying on a particular issue is not their core business. This is good news as it implies that top-down mapping might be much less sensitive to influence and mobilization bias than usually expected. One of the reasons is that one of the key sources of information on lobbying in the top-down mapping process entails using media reports to identify organizations with interests in a particular proposal. Using media reports for this purpose is not uncommon in studies of the influence of interest organizations (Andrews & Caren 2010; Barker-Plummer 2002; Bernhagen & Trani 2012; Binderkrantz 2012; Danielian 1994; De Bruycker & Beyers 2014). Yet, the media might well cite an organization as an example of how a particular proposal might bear on the activities of those it is trying to influence without that organization actually lobbying on the proposal. Also, an organization might comment on a proposal without it then doing anything substantive to influence its fate. Another reason is that a top-down 26 mapping, if conducted properly, results in a varying mix of highly prominent cases on which a large number of organizations are lobbying and a set of much less prominent cases where only a handful lobbyists are active. Many of the organizations identified in such a way are ‘followers’ who jump on the bandwagon that is led by a rather small set of leading organizations. Being part of the pack is crucial for generating necessary visibility and resources. 27 References Anderson, Jennifer, Adam Newmark, Virginia Gray, and David Lowery. 2004. “Mayflies and Old Bulls: Demographic Volatility and Experience in State Interest Communities,” State Politics and Policy Quarterly, 4 (2): 140-160. Andrews, K. T., & Caren, N. 2010. “Making the News Movement Organizations, Media Attention, and the Public Agenda.” American Sociological Review, 75, 841-866. Balme, Richard & Didier Chabanet. 2008. European Governance and Democracy: Power and Protest in the EU. Lanham, MD: Rowman & Littlefield,. Barker-Plummer, B. 2002. “Producing public voice: Resource mobilization and media access in the National Organization for Women.” Journalism & Mass Communication Quarterly. 79: 188-205. Baumgartner, Frank F. & Beth L. Leech. 1998. Basic Interests: The Importance of Groups in Politics and in Political Science. Princeton: Princeton University Press. Baumgartner, Frank R. and Beth L. Leech. 2001. “Interest Niches and Policy Bandwagons: Patterns of Interest Group Involvement in National Politics.” Journal of Politics. 63: 1468-2508. Baumgartner, Frank R., Jeffrey M. Berry, Marie Hojnacki, David C. Kimball, and Beth L. Leech. 2009. Lobbying and Policy Change. Chicago: University of Chicago Press. Berkhout, Joost and David Lowery. 2011. “Short-Term Volatility in the EU Interest Community.” Journal of European Public Policy. 18 (1): 1-16. Berkhout, Joost. 2010. Political Activities of Interest Organizations: Conflicting Interests, Converging Strategies. Mostert and van Onderen: Leiden. Bernhagen, P., and Trani, B. 2012. “Interest group mobilization and lobbying patterns in Britain: A newspaper analysis.” Interest groups & Advocacy. 1: 48-66. Beyers, J., A. Duer, D. Marshall and A. Wonka. 2014b. “Policy-Centred Sampling in 28 Interest Group Research: Lessons from the INTEREURO-project.” Interest Groups & Advocacy. 3: 160-173. Beyers, J., C. Braun, D. Marshall and I. De Bruycker. 2014c. “Let’s talk! On the practice and method of interviewing policy experts.” Interest Groups & Advocacy. 3: 174187. Beyers, J., R. Eising and W. Maloney. 2008. “Researching Interest Group Politics in Europe and Elsewhere: Much We Study, Little We Know?” West European Politics. 31: 1103-1128. Beyers, Jan, Laura Chaques Bonafant, Andreas Dur, Rainer Eising, Danica Fink-Hafner, David Lowery, Christine Mahoney, William Maloney, and Daniel Naurin. In 2014a. “The INTEREURO Project: Logic and Structure.” Interest Groups and Advocacy. 3 (2): 126-140. Binderkrantz, Anne S. 2012. “Interest groups in the media: Bias and diversity over time.” European Journal of Political Research. 51: 117-139. Binderkrantz, Anne S., Peter M. Christiansen & Helene H. Pedersen. 2014. “Interest Group Access to the Bureaucracy, Parliament, and the Media.” Governance. DOI: 10.1111/gove.12089 Bunea, A. and F. R. Baumgartner. 2014. “The state of the discipline: authorship, research designs, and citation patterns in studies of EU interest groups and lobbying.” Journal of European Public Policy. doi: 10.1080/13501763.2014.936483. Danielian, L.H., & Page, B.I. 1994. “The heavenly chorus: Interest group voices on TV news.” American Journal of Political Science. 38: 1056-1078. De Bruycker, Iskander and Jan Beyers. 2014. “Balanced or Biased? Interest Groups and Legislative Lobbying in the European News Media.” Political Communication. forthcoming. 29 Dür, Andreas and Gemma Mateo. 2013. “Gaining Access Or Going Public? Interest Group Strategies in Five European Countries.” European Journal of Political Research. 52: 660-686. Gray, Virginia and David Lowery. 1996. The Population Ecology of Interest Representation: Lobbying Communities in the American States. Ann Arbor, Michigan: University of Michigan Press. Gray, Virginia, David Lowery, and Jennifer Benz. 2013. Organized Interests and Health Policy in the American States. Washington D.C: Georgetown University Press. Greenwood, J. and J Dreger. 2013. “The Transparency Register: A European Vanguard of Strong Lobby Regulation Quest.” Interest Groups & Advocacy. 2: 139-162. Halpin, Darren and Anna S. Binderkrantz. 2011 “Explaining Breadth of Policy Engagement: Patterns of Interest Group Mobilization in Public Policy.” Journal of European Public Policy. 8 (2): 201-219. Halpin, Darren and Grant Jordan. 2012. “Estimating Group and Associational Populations: Endemic Problems but Encouraging Progress.” In Halpin, Darren and Grant Jordan (eds.) The Scale of Interest Organization in Democratic Politics: Data and Research Methods. Palgrave Macmillan, Basingstoke. Halpin, Darren, Baxter, G. & MacLeod, I. 2011. “Multiple Arenas, Multiple Populations: Counting Organized Interests in Scottish Public Policy.” In Halpin, Darren and Grant Jordan (eds.) The Scale of Interest Organization in Democratic Politics: Data and Research Methods. Palgrave Macmillan, Basingstoke. Halpin, Darren. 2011. “Explaining Policy Bandwagons: Organized Interest Mobilization and Cascades of Attention.” Governance, 24: 205-230. Hanagraaff, Marcel. 2014. All the World’s a Stage: Interest Group Politics at WTO and UNFCC Negotiating Conferences. Antwerp: University of Antwerp. 30 Hanegraaff, Marcel, Jan Beyers and Iskander De Brucyker. 2014. "In Front or Behind the Scene? The Political Strategies of Lobbyists at Global Diplomatic Conferences." under review. Heinz, John P., Edward O. Laumann, Robert L. Nelson, and Robert Salisbury. 1993. The Hollow Core. Cambridge, Massachusetts: Harvard University Press. Jentges, Eric, Matthias Brändli, Patrick Donges, and Offried Jarren. 2012. “Die Kommunikation Politischer Interessengruppen in Deutschland: Adressaten, Instrumente Und Logiken.” Studies in Communication|Media. 3/4: 281-409. Johnson, E.W. 2013. “Toward International Comparative Research on Associational Activity: Variation in the Form and Focus of Voluntary Associations in Four Nations.” Nonprofit and Voluntary Sector Quarterly. 43: 163S-181S. Jordan, Grant, Darren Halpin & William Maloney. 2004. “Defining Interests: Disambiguation and the Need for New Distinctions.” British Journal of Political Science. 6: 195-212. Jordan, Grant, Frank Baumgartner, John, McCarthy, Shaun Bevan, and Jamie Greenan. 2012. “Tracking Interest Group Populations in the US and the UK.” In Halpin, Darren and Grant Jordan (eds.) The Scale of Interest Organization in Democratic Politics: Data and Research Methods. Palgrave Macmillan, Basingstoke. Klüver, Heike, Christine Mahoney and Mark Opper. 2014. “Framing in context: How interest groups frame policy debates in the European Union.” under review. Klüver, Heike: Citizens, interest groups and legislative activity: A longitudinal study of issue evolution in Germany, unpublished manuscript Klüver, Heike. 2013. Lobbying in the European Union: Interest Groups, Lobbying Coalitions and Policy Change. Oxford: Oxford University Press. Leech, Beth L., Frank R. Baumgartner, Timothy La Pira, and Nicholas A. Semanko. 2005. 31 “Drawing Lobbyists to Washington: Government Activity and the Demand for Advocacy.” Journal of Politics. 58: 19-30. Lowery, David and Virginia Gray. 2004. “A neopluralist perspective on research on organized interests.” Political Research Quarterly. 57: 164-175. Lowery, David. 2007. “Why Do Organized Interests Lobby? A Multi-Goal, Multi-Context Theory of Lobbying.” Polity. 39 (1): 29-54. Mahoney, Christine. 2008. Brussels Versus the Beltway: Advocacy in the United State and the European Union. Washington, D.C.: Georgetown University Press. Salisbury, R.H. (1994) Interest Structures and Policy Domains: A Focus for Research. Crotty, W.J., M.A. Schwartz and J.C. Green (eds) Representing Interests and Interest Group Registration. , University Press of America Washington, DC: 12-20. Schattschneider, E. E. 1960. The Semisovereign People. New York: Holt, Rinehart, and Winston. Schlozman, Kay Lehman, Sidney Verba, and Henry E. Brady. 2012. The Unheavenly Chorus: Unequal Political Voice and the Broken Promise of American Democracy. Princeton: Princeton University Press. Sorurbakhsh, Laila. 2014. “Population Ecology and European Interest Groups Over Time.” European Political Science .13: 61-77. Thrall, A.T. 2006. “The Myth of the Outside Strategy: Mass Media News Coverage of Interest Groups.” Political Communication. 23: 407-420. Toshkov, Dimiter, David Lowery, Brendan Carroll, and Joost Berkhout. 2013. “Timing is Everything? Organized Interests and the Timing of Legislative Activity.” Interest Groups and Advocacy. 2 (1): 48-70. Walker, Jack L. 1991. Mobilizing Interest Groups in America. Ann Arbor Michigan: University of Michigan Press. 32 Webb, Susan. 2006. “Sweet-Talking the Fourth Branch: Assessing the Influence of Interest Group Comments on Federal Agency Rulemaking.” Journal of Public Administration Research and Theory. 26: 103-124. Witko, Christopher. In Press. “Case Study Approaches to Studying Organization Survival and Adaptation.” In David Lowery, Darren Halpin, and Virginia Gray, eds. The Organizational Ecology of Interest Communities Reconsidered. Palgrave Macmillan, Basingstoke. Wonka, Arndt, Frank R. Baumgartner, Christine Mahoney, and Joost Berkhout. 2010. “Measuring the Size and Scope of the European Union Interest Group Population.” European Union Politics. 11: 463-476. 33 Table 1. Number of groups identified in EC-expert interviews, media sources and interest group interviews N legislative proposals Total number of groups identified 95 EC-expert interviews Media sources; mentions Media sources; statements Total number identified 125 In 141 Interest group interviews 85 70 125 658 288 379 291 1496 Figure 1. Venn-diagram bottum-up and top-down populations Venn-diagram (N=2895) Bottum-up population N=1960 (68%) Overlap N=347 (12%) Top-down population N=588 (20%) 34 35 36 37 38 39 40 41 42 43
© Copyright 2026 Paperzz